DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation
Abstract
DeepPresenter is an agentic framework for presentation generation that adaptively plans and refines slide artifacts through environment-grounded reflection, achieving state-of-the-art performance with reduced computational costs.
Presentation generation requires deep content research, coherent visual design, and iterative refinement based on observation. However, existing presentation agents often rely on predefined workflows and fixed templates. To address this, we present DeepPresenter, an agentic framework that adapts to diverse user intents, enables effective feedback-driven refinement, and generalizes beyond a scripted pipeline. Specifically, DeepPresenter autonomously plans, renders, and revises intermediate slide artifacts to support long-horizon refinement with environmental observations. Furthermore, rather than relying on self-reflection over internal signals (e.g., reasoning traces), our environment-grounded reflection conditions the generation process on perceptual artifact states (e.g., rendered slides), enabling the system to identify and correct presentation-specific issues during execution. Results on the evaluation set covering diverse presentation-generation scenarios show that DeepPresenter achieves state-of-the-art performance, and the fine-tuned 9B model remains highly competitive at substantially lower cost. Our project is available at: https://github.com/icip-cas/PPTAgent
Community
Open Sourced at: https://huggingface.co/collections/ICIP/deeppresenter
Project: https://github.com/icip-cas/PPTAgent
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- Code2World: A GUI World Model via Renderable Code Generation (2026)
- TermiGen: High-Fidelity Environment and Robust Trajectory Synthesis for Terminal Agents (2026)
- SceneReVis: A Self-Reflective Vision-Grounded Framework for 3D Indoor Scene Synthesis via Multi-turn RL (2026)
- Code2Worlds: Empowering Coding LLMs for 4D World Generation (2026)
- MUSE: A Multi-agent Framework for Unconstrained Story Envisioning via Closed-Loop Cognitive Orchestration (2026)
- ANCHOR: Branch-Point Data Generation for GUI Agents (2026)
- RAVEL: Reasoning Agents for Validating and Evaluating LLM Text Synthesis (2026)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Slide 1 — Title
KRISHIMITRA AI
AI-Powered Agriculture & Direct Farmer-to-Buyer Marketplace
“Grow Smarter. Sell Directly. Earn Better.”
Project Team: [pixel punks]
BCA – Computer Applications
College :- GSS Belagavi
Slide 2 — Problem Statement
THE PROBLEM
Farmers often depend on multiple middlemen.
A significant portion of the final selling value may go through intermediaries.
Farmers have limited direct access to companies and bulk buyers.
Difficulty in deciding which crop to grow.
Crop diseases can reduce yield and income.
Agricultural information is not easily accessible to every farmer.
Traditional Flow:
Farmer → Trader → Wholesaler → Distributor → Company/Consumer
Slide 3 — Our Solution
KRISHIMITRA AI — OUR SOLUTION
A single digital platform connecting:
FARMER ↔ AI AGRICULTURAL ASSISTANCE ↔ MARKETPLACE ↔ BUYER/COMPANY
Core Features
AI Crop Recommendation
AI Crop Disease Detection
AI Agricultural Assistant
Direct Farmer-to-Buyer Marketplace
Product Image Upload & Management
Quantity-Based Ordering
Multilingual Support
Farmer & Company Profiles
Slide 4 — Farmer Features
FARMER MODULE
👨🌾 Farmer can:
Create and manage profile
Add farm details
Enter soil, water and land information
Get AI crop recommendations
Upload crop image for disease detection
Ask agricultural questions to AI assistant
Upload agricultural products
Add product image, quantity and price
Receive buyer orders
Manage listed products
Slide 5 — Direct Marketplace
FARMER-TO-BUYER MARKETPLACE
Without KrishiMitra
Farmer → Middleman → Wholesaler → Company
With KrishiMitra
Farmer → KrishiMitra → Company/Buyer
Farmer
Upload product image
Product name
Available quantity
Expected price
Product details
Buyer
Search products
View farmer & product
View product image
Select required quantity
Confirm order
Goal: Better market access and reduced unnecessary intermediary dependency.
Slide 6 — AI Crop Planner
AI CROP RECOMMENDATION
Input:
Location
Soil type
Soil pH
Water availability
Land area
Budget
Previous crops
Season
Soil report (optional)
↓
Python + Machine Learning Model
↓
Output:
Suitable crop recommendations
Soil compatibility
Water requirement
Growing season
Basic farming guidance
Example:
Rice → Maize → Groundnut
Slide 7 — AI Disease Detection
AI CROP DISEASE DETECTION
📷 Upload / Capture Crop Image
↓
Image Processing
↓
AI / Computer Vision Model
↓
Disease Classification
↓
Result
Detected crop/plant
Possible disease
Confidence score
Symptoms
General preventive guidance
Recommended next steps
Technologies: Python + OpenCV/Pillow + ML/DL
Slide 8 — AI Agricultural Assistant
AI AGRICULTURAL ASSISTANT
Farmers can ask:
What crop should I grow?
What disease is affecting my crop?
What conditions does this crop need?
What should I do after disease detection?
How can I improve my farming practices?
How can I sell my products?
Multilingual Support
English | Kannada | Hindi | Marathi
Future
Voice Input
Voice Output
WhatsApp AI Assistant
Slide 9 — Technology Stack
TECHNOLOGY STACK
Frontend
Next.js
UI
Routing
Dashboard
Marketplace
API Integration
TypeScript
Type safety
Reliable and maintainable code
JavaScript
Interactive UI
Client-side logic
Backend / Database
PocketBase
Authentication
Database
File/Image Storage
APIs
User Management
AI / ML
Python
AI & data processing
Scikit-learn
Machine Learning
Training
Classification
Prediction
OpenCV / Pillow
Image processing
Slide 10 — System Architecture
KRISHIMITRA AI — SYSTEM ARCHITECTURE
FARMER / BUYER
↓
NEXT.JS APPLICATION
TypeScript + JavaScript
↓
POCKETBASE
┌────────┼────────┐
↓ ↓ ↓
Users Products Orders
↓ ↓ ↓
Images Farmer Data
↓
PYTHON AI SERVICES
↓
┌─────────┴─────────┐
↓ ↓
Crop Recommendation Disease Detection
↓ ↓
└─────────┬─────────┘
↓
AI RESULT / GUIDANCE
Slide 11 — Complete User Flow
FROM FARM TO MARKET
Register
↓
Create Farmer Profile
↓
Enter Farm & Soil Details
↓
AI Crop Recommendation
↓
Grow Crop
↓
AI Disease Detection
↓
Harvest
↓
Upload Product + Image
↓
Buyer Discovers Product
↓
Buyer Selects Required Quantity
↓
Confirm Order
↓
Direct Procurement
KrishiMitra connects the agricultural lifecycle from decision-making to selling.
Slide 12 — Impact & Future Scope
IMPACT & FUTURE SCOPE
Expected Impact
Better agricultural decision-making
Earlier disease identification
Improved market access
Direct connection with buyers
Transparent product information
Reduced unnecessary intermediary dependency
Future Scope
AI yield prediction
Crop price prediction
Weather alerts
IoT soil sensors
Smart irrigation
Digital payments
Logistics integration
Government scheme integration
Voice-based agricultural assistant
WhatsApp integration
Bulk procurement & bidding
Slide 13 — Conclusion
KRISHIMITRA AI
“Grow Smarter. Detect Earlier. Sell Directly.”
KrishiMitra combines:
AI + Agriculture + Marketplace + Technology
to help farmers:
🌱 Make Better Farming Decisions
📷 Detect Crop Problems Earlier
🏪 Reach Buyers Directly
💰 Improve Market Opportunities
Technology
Next.js + TypeScript + JavaScript + PocketBase + Python + Scikit-learn + Computer Vision
THANK YOU
Questions & Answers
Get this paper in your agent:
hf papers read 2602.22839 Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 2
Datasets citing this paper 1
Spaces citing this paper 0
No Space linking this paper